Time and the plan

How long a plan runs changes what its numbers mean, not just how many of them there are. This page covers the horizon the plan simulates to, why it's set where it is, how a short plan and a long one behave differently under the same assumptions, and the conventions the engine uses for "today" and "a year" along the way.

Why the plan runs to 95

The plan runs from today to the primary member's horizon age — 95 by default, and adjustable from 60 to 120 on their card in the Household section (see Households and the plan).

Ninety-five is a deliberately long horizon, not a best guess at how long anyone in your household will actually live. The risk a retirement plan has to answer for is running out of money too soon, not dying too soon — the simulation has no mortality model at all (see What the model leaves out), so nothing in it ever "dies," and a shorter life costs the plan nothing it can see. Outliving your money is the failure mode a plan can actually miss, and cutting the horizon to a more average age would only hide that risk by ending the projection before it has a chance to show up. If your household has a specific reason to plan to a different age, the field is right there — the point of 95 as a default is only that a shorter one would understate the risk, never that it would overstate it.

Fifteen years vs. thirty-plus

The same assumptions play out differently depending on how many years the plan has to cover.

Wider bands

Uncertainty compounds year over year, so the gap between an unlucky and a lucky simulated path grows the farther out you look — a plan running to 95 fans out wider at the far end than the same plan cut off at 80 would.

More sensitive to the return assumption

A return you're off by half a point barely shows up over five years; compounded over thirty, it can be the difference between two very different-looking plans (see Default assumptions).

Recovery time comes from working years, not from a longer horizon

It's tempting to think a longer plan gives a bad early decade more time to be made up — but in this engine, simply extending the horizon age mostly doesn't do that. Ruin here is absorbing: a path that ever hits zero has failed permanently, whatever a later decade would have done for it (see Monte Carlo and scenarios). Running the simulation for more years past retirement only appends more retired, spending-not-saving years to it — exposure, not runway — so pushing the horizon out on its own can hold your success rate flat or lower it, never rescue an early loss.

The real recovery lever is how many years you're still contributing. A downturn that hits while you're still working has decades of ongoing saving and growth ahead of it to close the gap; the same downturn hitting after you've retired and started drawing down doesn't have that lever, no matter how far out your horizon age is set. A longer working period — not simply a longer plan — is what buys a bad early decade room to recover in.

That last point comes with a catch that neither horizon length nor how long you work fixes on its own: the years right around your retirement date carry outsized weight, whatever the plan's total length. A downturn while you're still adding money behaves very differently from the same downturn just after you've started drawing on a portfolio that has stopped growing — it's the order returns arrive in, not only their average, that decides whether a given path survives. This is sequence-of-returns risk, and Monte Carlo and scenarios has a worked example of exactly how it plays out — two retirees who live through the same three years of market returns in a different order and end up $50,000 apart, despite identical averages.

Today's dollars

Every number the plan shows you — the fan chart, the year-by-year table, the success percentage — is stated in today's purchasing power, not the dollars that will actually be printed on anything in that future year. Rather than inflating decades of cash flows forward and asking you to mentally deflate them back, the engine deflates once, up front, and simulates in real terms the whole way through (see How the plan is simulated for the arithmetic).

So a 2055 row that shows a $60,000 spend doesn't mean you'll write checks for $60,000 that year — it means whatever it costs in 2055 to buy what $60,000 buys today. If the plan holds together at $2.1M in a 2055 row, that $2.1M is already stated in money you understand the value of, not a number inflated to look larger than it means.

Calendar conventions

The plan stores only a birth year for each person, not a birthday, so it treats every birthday as landing on January 1st. That sets the Social Security claiming boundary and the age required minimum distributions start — both switch on the January of the calendar year you'd turn that age, not on your actual birthday. It does not set the 59½ early-withdrawal-penalty boundary, which is the one age the plan tracks by month rather than by calendar year: it lifts in July of the year the owner turns 59 — a January 1st birthday plus fifty-nine and a half years, not January of that year.

None of this is free of consequence. Depending on where your real birthday falls in the year, the January-1st convention can move a claim, a required distribution, or the 59½ boundary by up to about a year in either direction — which changes real dollars, not just bookkeeping (see What the model leaves out for the direction of that error).

Within a year, not everything lands the same way either. Required minimum distributions and Roth conversions happen once, in the first simulated month of their calendar year — the plan's own first, partial year included. Spending, goals, health insurance, income and saving are smoothed evenly across the twelve months of the year they fall in, rather than landing all at once.

The fan chart and the Year by year table don't treat that first, partial year the same way, though, and the two can genuinely disagree in their first row because of it. The 1,000 simulated paths only charge the months actually left in the current year — starting the plan in September charges four twelfths of that year's figures, because you only live the months that are left. The Year by year table always shows a whole calendar year, starting from January of the current year regardless of what month it is when you're reading it, so its arithmetic reconciles by hand.

Retirement itself begins at the start of January of the year the household retires (the year the last working adult stops), and the plan's own final year ends at the start of January the primary member reaches their horizon age.